<jats:title>Abstract</jats:title> <jats:p>A high-resolution atmosphere–sea ice model is used to investigate the interactions between cyclones and sea ice cover in polar regions. For this purpose, a cyclone passage observed during the 1999 Fram Strait Cyclone Experiment (FRAMZY) is simulated for two consecutive days. The results of the coupled mesoscale transport and stream model–mesoscale sea ice model (METRAS–MESIM) are compared with aircraft and ice drift measurements. With the exception of temperature, all atmospheric parameters are well simulated. Main reasons for discrepancies were found in large differences between the measurements and the forcing data taken from the results of the regional model (REMO). In addition, advection was slightly wrong as a result of a 17° deviation in wind directions. The altogether well simulated wind field is interactively used to force the sea ice model MESIM; results agree well with drift buoy measurements. Average deviations of simulated and measured ice drift are smaller than 8° for direction and smaller than 3.7 cm s−1 for speed, which is less than 10% of the average speed. The simulated ratio between ice drift and wind velocity increases slightly during cyclone passage from 2.6% to 2.9%, a tendency also known from observations. During a 36-h period, the simulated sea ice concentration locally decreases up to 20% in accordance with measurements. A neglect of changing sea ice cover causes a decrease of the heat flux to the atmosphere from 53 to 12 W m−2. The values correspond to averages over the evaluation region (approximately 228 000 km2) and period (36 h). Temperature and humidity are decreased by 2 K and 0.2 g kg−1, respectively, over the ice-covered region. In contrast, the effect on pressure and wind remains small, probably because the cyclone does not move in the vicinity of the ice edge.</jats:p>